Text-mining-with-Linked-Life-Data

Download Report

Transcript Text-mining-with-Linked-Life-Data

Information Extraction
with
Linked Life Data
www.linkedlifedata.com
19/04/2011
Linked Life Data Overview
• Linked Life Data is a public
and free RDF warehouse
service
• Offers a SPARQL endpoint
• Integrates more than 25
popular biomedical data
sources
• Specifies many cross data
sources semantic mappings
• Exposes massive amounts of
linked data
#2
Semantic Annotations with LLD
#3
Simple Knowledge Organisation Schema (SKOS)
• SKOS is a data model
prefLabel
Respiratory Disease
• Serialized as RDF graph
• Published on the web in
a to be shared between
applications
• Efficient structuring of
terms in thesauri
• Allows the identification
of the concepts by URLs
broader
Bronchial Asthma
broader
altLabel
broader
prefLabel
prefLabel
Asthma
Chronic Obstructive Asthma
#4
SKOS Semantics to Map Across Different Thesauri
Relationship
Semantics
Exact match
Transitive equivalence
Close match
Equivalent only for
search purposes
Broader match
Generalization of a
concept
Narrower
match
Specialization of a
concept
Related
Unspecified relation
(no real semantics)
Example
Inverse of broader match
#5
http://linkedlifedata.com/resource/umls/id/C0004096
http://linkedlifedata.com/resource/umls/id/C0004096
Different formats
Concept definition
Concept types
Concept labels
Querying LLD web interface with SPARQL
• LLD supports a database like
web interface for executing
SPARQL queries, where:
• You can test you query
• Populate directly from the
database your gazetteer
(e.g. LKB Gazetteer)
#8
Querying SKOS Data
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX lld: <http://linkedlifedata.com/resource/>
Namespace prefixes
SELECT DISTINCT ?label ?concept ?top
WHERE {
?top skos:prefLabel "Respiration Disorders".
concept with this name
?concept skos:broader ?top.
child concepts
?concept skos:inScheme lld:umls.
part of UMLS
?concept rdfs:label ?label.
all their synonyms
}
Return all “Respiration Disorder” concepts in LLD and all their IDs and labels
#9
How to use LLD from Gate Developer?
• Use LKB Gazetteer part of Gate Developer’s distribution
• Modify “query.txt” file with the results you have tested
• Point the gazetteer to the “repositoryURL” of:
http://linkedlifedata.com/sparql
#10
Conclusion
#11